Model

Load and manage ONNX Runtime GenAI models

The Model class represents a loaded ONNX Runtime GenAI model. It is the primary entry point for loading models and creating other components like tokenizers and generators.

Constructor

Model(string modelPath)

Creates a new Model instance from a model directory.

modelPath · string · required

Path to the directory containing the model files

Model.cs:14-17

using Microsoft.ML.OnnxRuntimeGenAI;

using Model model = new Model("/path/to/model");

Model(Config config)

Creates a new Model instance from a configuration object.

config · Config · required

Configuration object with model settings and provider options

Model.cs:19-22

using Microsoft.ML.OnnxRuntimeGenAI;

using Config config = new Config("/path/to/model");
config.AppendProvider("cuda");
using Model model = new Model(config);

Methods

GetModelType()

Returns the type of the model (e.g., “phi3v”, “phi4mm”, “qwen2_5_vl”).

Returns: string - The model type identifier

Model.cs:26-38

string modelType = model.GetModelType();
Console.WriteLine($"Model type: {modelType}");

Dispose()

Releases the resources used by the model. Models implement IDisposable and should be disposed when no longer needed.

model.Dispose();
// Or use 'using' statement for automatic disposal
using Model model = new Model("/path/to/model");

Complete Example

Here’s a complete example of loading a model and using it with a tokenizer and generator:

using Microsoft.ML.OnnxRuntimeGenAI;

string modelPath = "/path/to/model";

// Load model with default configuration
using Model model = new Model(modelPath);

// Get model type
string modelType = model.GetModelType();
Console.WriteLine($"Loaded model type: {modelType}");

// Create tokenizer from model
using Tokenizer tokenizer = new Tokenizer(model);

// Create generator params
using GeneratorParams generatorParams = new GeneratorParams(model);
generatorParams.SetSearchOption("max_length", 1024);

// Create generator
using Generator generator = new Generator(model, generatorParams);

// Model is automatically disposed when leaving the 'using' scope

Example with Custom Configuration

Load a model with custom execution provider settings:

examples/csharp/ModelChat/Program.cs:586-590

using Microsoft.ML.OnnxRuntimeGenAI;

string modelPath = "/path/to/model";
string executionProvider = "cuda";

// Create configuration
using Config config = new Config(modelPath);
config.ClearProviders();
config.AppendProvider(executionProvider);

// Create model from configuration
using Model model = new Model(config);
Console.WriteLine($"Model loaded with {executionProvider} provider");

See Also

• Tokenizer - Create tokenizers from models
• Generator - Generate text with models
• GeneratorParams - Configure generation parameters